87

Chemical proteomics and its application to drug discovery Douglas A Jefferyà and Matthew Bogyoy

The completion of the human genome sequencing project has For this reason, the discipline of biochemistry has provided a ¯ood of new information that is likely to change the become an increasingly important component in efforts to way scientists approach the study of complex biological reap maximum reward from the information provided by systems. A major challenge lies in translating this information into genomics. The term `proteomics' has been coined to new and better ways to treat human disease. The de®ne the study of protein biochemistry on a genome multidisciplinary science of chemical proteomics can be used to wide scale. distill this ¯ood of new information. This approach makes use of synthetic small molecules that can be used to covalently The broad ®eld of proteomics has as its primary goal the modify a set of related and subsequently allow their understanding of the structure, function, expression, cel- puri®cation and/or identi®cation as valid drug targets. lular localization, interacting partners, and regulation of Furthermore, such methods enable rapid biochemical analysis every protein produced from a complete genome [5,6].At and small-molecule screening of targets thereby accelerating the the heart of this ®eld are many new technologies and often dif®cult process of target validation and drug discovery. techniques that are being developed to facilitate this daunting task. This review focuses on the multidisciplin- Addresses ary ®eld of chemical proteomics, which makes use of Celera Genomics, 180 Kimball Way, South San Francisco, CA 94080, synthetic organic chemistry, cell biology, biochemistry, USA and mass spectrometry. A central component of this ®eld à e-mail: [email protected] is the design of speci®c protein-modifying reagents that ye-mail: [email protected] can be used for functional studies of distinct families within a complex proteome. These chemical Current Opinion in Biotechnology 2003, 14:87±95 probes are designed to covalently modify a target enzyme in such a way that it can be subsequently identi®ed and/or This review comes from a themed issue on Analytical biotechnology puri®ed [7]. Through synthetic organic chemistry, these Edited by Norm Dovichi and Dan Pinkel probes can be designed to react with mechanistically or functionally distinct families or subfamilies of enzymes. 0958-1669/03/$ ± see front matter ß 2003 Elsevier Science Ltd. All rights reserved. In the cases where enzymes utilize a nucleophilic attack mechanism, probes can be designed to modify speci®c DOI 10.1016/S0958-1669(02)00010-1 active-site residues in a manner that requires enzymatic activity of the target. Such probes have been termed Abbreviations activity-based probes (ABPs) to re¯ect their need for ABP activity-based probe an active enzyme to facilitate covalent modi®cation. AFPB af®nity-based probe Other chemical probes have been designed that target 50-FSBA 50-¯uorosulfonyl benzoyl adenosine non-catalytic residues on and enzymes. These ICAT isotope-coded af®nity tag SDS±PAGE sodium dodecyl sulfate±polyacrylamide gel af®nity-based probes (AFBPs) require highly selective, electrophoresis tight binding to targets to be useful probes for distinct protein/enzyme families. Still other chemical probes have been designed to modify proteins en masse for quantitative Introduction studies of complete proteomes. Regardless of their The completion of the human genome has created much mechanism of action, chemical probes are ®nding increas- excitement from the impact that this wealth of informa- ing use in the ®eld of proteomics and have great potential tion is likely to have on the process of drug discovery and to aid in the process of target identi®cation, target valida- development [1±3]. It has been postulated that scientists tion, and drug discovery. could use genome information to identify and validate a host of new drug targets and tailor speci®c drugs based on Anatomy of a chemical probe an individual's detailed genetic makeup [4]. Although In their most basic form, chemical probes consist of three this new ®eld of genomics holds much promise, it is clear distinct functional elements (Figure 1): a reactive group that analysis of DNA or RNA content alone is not suf®- for covalent attachment to the enzyme; a linker region cient to understand cell biology and disease. Further- that can modulate reactivity and speci®city of the reactive more, the estimated 30 000±40 000 protein-encoding group; and a tag for identi®cation and puri®cation of genes in the human genome could result in 10±100 times modi®ed enzymes. We outline the chemical structures this number of unique proteins through post-transcrip- of each of these reactive groups and highlight speci®c tional and post-translational processing and modi®cation. examples of how they are used in chemical probes. www.current-opinion.com Current Opinion in Biotechnology 2003, 14:87±95 88 Analytical biotechnology

Figure 1

Activity-based probes (ABPs)

Label/tag Linker/specificity Reactive group [Ref] O O HN NH O O

S Alkyl O O Epoxide [16] Affinity (Biotin)

S O O O O F F O F B P N N O Polyethylene glycol (PEG) Fluorophosphonate [10] Fluorescent (Bodipy) R1 O R3 O O H S 125 N I O N H O N HO O R2 O NO 2 Peptide Radioactive Sulfonate [25]

Current Opinion in Biotechnology

Structure of a chemical probe. A chemical probe has three basic components: a reactive group for covalent attachment to the enzyme of interest; a linker region to provide spacing and specificity; and a tag to allow for identification and/or purification. Specific examples of each are shown with references where relevant. See text for details.

Structure of the reactive group inhibitors provide a rich source of reactive groups that Perhaps the greatest challenge in the design of a chemical have been designed based on subtle differences in reac- probe is selection of a reactive group that provides the tion mechanisms for the major protease families. Not necessary covalent modi®cation of a target protein. The surprisingly, many of the best examples of chemical dif®culty lies in the duality of this functional group as it probes have been designed to target proteolytic enzymes must be both reactive towards a speci®c residue on a [8±12] (for review see [7]). protein and inert towards other reactive species within the cell or cell extract. In general, the reactive groups of Most reactive groups take the form of suicide/mechan- most of the successfully designed chemical probes have ism-based inhibitors or af®nity labeling reagents. These been based on the chemistries of covalent, mechanism- have been used for decades as drugs (e.g. aspirin), as tools based inhibitors of various enzyme families. Such inhi- to identify the active-site residues in enzymes, and as bitors rely on mechanistic differences of individual tools to understand the mechanism and function of enzyme classes as a means for selective targeting. For enzyme in vivo [13,14]. There are far too many example, and cysteine proteases utilize a catalytic examples of mechanism-based inhibitors or af®nity amino acid nucleophile in the , yet each has a labeling reagents to cover within the scope of this review, different nucleophilic residue and a distinct catalytic so we use speci®c examples to highlight the general types mechanism allowing for the design of chemistries that of reagents that have potential applications in probe react with one class and not the other. In fact, protease design.

Current Opinion in Biotechnology 2003, 14:87±95 www.current-opinion.com Chemical proteomics in drug discovery Jeffery and Bogyo 89

Figure 2

Probe type Mechanism Examples [Ref]

- HN N H HN N H S O S I: Mechanism-based O DCG-04 [16] OEt ABP OEt Tag N FP-Biotin [10] Tag N H H O OH O O

-S - O- O O O NH -O -O NH O P S O O P O O O O HO NH O II: Suicide NH F O F H DFPP [18,19] ABP NH Sulbactam [17] O F O N H NH H H N HN O HN O

Tag Tag

NH 2 NH2 O O N O N O N O S O S N N III: Affinity alkylating F O O N NH O N NH O N 5′-FSBA [21] AFBP 2 HO OH HO OH

O O I Tag N Tag N H H IV: General alkylating HS S ICAT [24]

Current Opinion in Biotechnology

Mechanism of action of the four primary classes of chemical probes. Examples of the four general types of reactive groups that can be incorporated into chemical probes with the mechanism of action of an example reactive group for each shown in the middle column. The right-hand column contains specific examples of chemical probes/inhibitors that belong to each class with the relevant references listed.

There are four general classes of reactive groups that have active. Probes carrying this class of reactive group must been used to design chemical probes (Figure 2). The ®rst rely on the selectivity of the probe scaffold to direct two classes (type I and type II) are ABPs based on true modi®cation to speci®c enzyme/protein classes. The ®nal mechanism-based or suicide inhibitors and require active class of reactive groups used for probe design (type IV) target enzymes to react with the probe. In the case of the contains non-speci®c alkylating groups that react with mechanism-based probes (type I), the key nucleophile is targets based only on the intrinsic reactivity of a speci®c the catalytic residue of the enzyme normally involved in amino acid residue such as cysteine. This class of probes attack of a substrate. This type of probe can therefore be has recently been shown to be valuable for bulk proteo- made selective based on knowledge of catalytic mechan- mic analysis using mass spectrometry methods [15]. ism and is often tailored to the reactivity of the speci®c nucleophilic atom used by the enzyme class (i.e. a sulfur Examples of type I reactive groups in chemical probes used by cysteine proteases). The second class of ABPs include peptide acyloxymethyl ketones and epoxides that (type II) contain a `masked electrophile' that is uncovered have been shown to ef®ciently and selectively label after the probe functions as substrate for the target cysteine proteases of the caspase and papain families, enzyme. The unmasked electrophile is able to react with respectively [11,16] (Figure 2). These reactive groups nearby, non-catalytic, nucleophilic residues in the active make use of an electrophilic carbon that is susceptible site. Af®nity alkylating probes (AFBPs, type III) contain to attack by the active-site nucleophile of the enzyme. In af®nity-based labeling groups that require only a strong the case of an epoxide probe the result is the formation of nucleophile or electrophile in the vicinity of the active- a covalent bond between the active-site thiol residue and site pocket and do not require the enzyme to be fully the electrophilic carbon in the epoxide ring (Figure 2). In www.current-opinion.com Current Opinion in Biotechnology 2003, 14:87±95 90 Analytical biotechnology

theory, any reactive group that mimics a substrate and has vated by UV light or chemical reduction can be used as an electropositive carbon could be used to covalently af®nity labels for GTP-binding proteins [23]. The poten- modify the active-site residue of an enzyme that uses a tial exists for the use of af®nity labeling reagents to nucleophilic attack mechanism (e.g. through cysteine or modify a wide variety of enzymes that bind small mole- serine residues). Furthermore, many other enzyme classes cule substrates (such as ATP) or cofactors. (e.g. , DNA repair enzymes, ) utilize a nucleophilic attack mechanism at some point during the The ®nal class of reactive groups (type IV) makes use of a process of catalysis. To date, type I reactive groups are the general alkylating reagent. The iodoacetamide group of most commonly used in ABPs and this is likely to continue the isotope-coded af®nity tag (ICAT) [24] is designed to to be the case as new probes are developed for other react with free sulfhydryl groups on cysteine residues enzyme classes. (Figure 2). The result is the covalent attachment of the probe to all free sulfhydryls on a protein or in a protein Two examples of type II mechanism-based inhibitors are mixture. This reactive group has been shown to be very sulbactam, a b-lactamase inhibitor used in combination useful in the quantitation of relative bulk protein levels in with other antibiotics to combat bacterial antibiotic resis- two different proteomes and is reviewed elsewhere [15]. tance [17], and DFPP, a probe that has been used to alkylate protein phosphatases [18,19]. The proposed Structure of the linker region mechanism of action against phosphatases for DFPP The linker region of a chemical probe connects the reac- involves initial attack by the active-site nucleophile on tive group to the tag used for identi®cation and/or puri®ca- the phosphotyrosine mimetic resulting in dephosphoryla- tion (Figure 1). The linker region can serve multiple tion of the probe. This leads to production of a reactive purposes. Its primary function is to provide enough space quinone methide that can react with a nucleophilic side- between the reactive group and the tag to prevent steric chain, lysine or cysteine for example, found in the active hindrance that could block access of the reactive group or site, resulting in irreversible alkylation of the enzyme. accessibility of the tag for the purpose of puri®cation. This One drawback to this type of reactive group is that it can is often accomplished using a long-chain alkyl or poly- lead to diffusion of unmasked electrophiles from the ethelene glycol (PEG) spacer. The alkyl linker can be active-site pocket and alkylation of other sites on the particularly useful to modulate hydrophobicity and allow target enzyme or other proteins that carry nucleophilic entry into live cells or tissues, whereas the PEG linker can residues on their surface. This in fact is the mechanism of confer more solubility to hydrophobic probes in aqueous action of a similar probe, ortho-(di¯uoromethyl)aryl-b-D- solutions. glucoside, that can alkylate b-galactosidase without inhi- biting its enzymatic activity [20]. This problem can The linker can also incorporate speci®city elements used potentially be overcome by adding speci®city elements to target the probe to a desired enzyme or family of to the probe (see below) to increase its af®nity for the enzymes. To date, these speci®city elements normally active site of the enzyme and keep the unmasked elec- take the form of a peptide or peptide-like structure, trophile bound long enough for speci®c alkylation of the particularly for the ABPs used that target proteases (see desired enzyme to occur. below) where two to four amino acids are incorporated chemically into the probe to provide speci®c binding to Type III AFBPs differ from ABPs (type I and II) in that protease active sites. Yet other examples of a chemical they do not require an active enzyme for modi®cation. In probes exist in which the main speci®city elements are most cases they are substrate analogs that contain a contained within the structure of the reactive group [25]. reactive center that is susceptible to attack by an electro- phile or nucleophile in the active site of the enzyme or The use of very large peptides or proteins to provide a that can be activated through the subsequent addition of high degree of target speci®city to a chemical probe also chemicals or UV light. An example of an af®nity-based holds much promise [26,27]. Recently, techniques have labeling reagent is the nucleotide analog 50-¯uorosulfo- been developed that can be used to modify a recombi- nylbenzoyl adenosine (50-FSBA) that has been used nantly expressed ubiquitin with an electrophilic reactive extensively to identify the active site of nucleotide-bind- group. The resulting protein probe can then be used to ing enzymes [21] (Figure 2). This ATP mimetic binds to covalently modify proteases that process ubiquitin in the active site of an enzyme bringing the reactive ¯uor- vitro. This method is particularly important because it osulfonyl group into close proximity to active-site nucleo- has been dif®cult to chemically synthesize speci®city philes such as lysine and cysteine. This results in elements for de-ubiquinating enzymes due to the need alkylation of the enzyme and loss of activity. 50-FSBA for a very long (>70 amino acids) stretch of amino acids. has been shown to be a potent inhibitor of protein kinases This technique of making ABPs from recombinant pro- through alkylation of the invariant active-site lysine [22], teins could be very useful in the design of probes for other as well as an inhibitor of many other nucleotide-binding enzymes or protein-binding domains that require sub- proteins [21]. Similarly, GTP analogs that can be acti- stantial protein recognition elements for speci®city.

Current Opinion in Biotechnology 2003, 14:87±95 www.current-opinion.com Chemical proteomics in drug discovery Jeffery and Bogyo 91

Structure of the tag to cellular localization and regulation. Here we discuss The purpose of the tag on a chemical probe is to allow the applications of chemical probes primarily in the quick and simple identi®cation and puri®cation of probe- context of drug discovery and development, but these modi®ed proteins. In general, the tag is the primary methods also have many important applications in basic element that distinguishes an ABP or AFBP from a research. stand-alone mechanism-based inhibitor. The most com- monly used tags are biotin, ¯uorescent, and radioactive Applications to target identi®cation and validation tags (Figure 1). Because the simplest and most cost- One of the primary challenges facing pharmaceutical effective method to separate proteins involves the use of companies is allocating their drug discovery resources protein gels, tags used in probe design must be compatible towards therapeutically relevant protein targets. The pro- with SDS±PAGE methods. Biotin facilitates detection by cess of target selection is complex and choosing a poor simple western blot approaches using a reporter avidin target can cost time and money, especially in competitive molecule in place of the standard secondary antibody. markets. It is estimated that each new drug candidate that Fluorescent and radioactive tags can be visualized by direct enters the clinic costs 250 full-time employee years and scanning of gels with a ¯uorescent scanner/phosphorima- $70 million [36]. Part of this cost stems from the high ger such as the Typhoon scanner from Amersham Bio- attrition rate of targets during the early stage of drug sciences [28]. Fluorescent and radioactive tags have several discovery, with an estimated sixty new targets required advantages over biotin. They are typically faster to use to generate three new marketed drugs every year [4]. because they involve minimal time and handling. They Chemical probes provide a method to focus initial new also are more sensitive and have a greater dynamic range target identi®cation efforts towards proteins that are more than streptavidin±biotin detection methods, especially easily validated and most likely to be effective drug when using the recently developed ¯uorescent tags such targets, thereby creating a higher potential for success. as the Alexa Fluors from Molecular Probes [29]. Finally, Drug targets that come from chemical proteomic screens they have the added advantage of allowing multiplexing of are pre-selected as susceptible to inhibition by small- samples based on non-overlapping excitation/emission molecule drugs by virtue of the method of their identi®ca- spectra of the tags. This allows one to use probes with tion. Chemical probes can also be used in validation different colored ¯uorescent tags in different experiments experiments in animal or cellular models both at early and readily obtain all results on a single gel [30,31]. stages of disease model selection and during initial target validation experiments. Additionally, reactive groups and Regardless of the bene®ts of ¯uorescent and radioactive speci®city elements of the chemical probe can serve as a tags, biotin remains the most commonly used tag because starting point for small-molecule inhibitor design. of its ability to provide both a gel-based method of detection and a method for puri®cation of labeled Chemical probes have already been used to identify enzymes on streptavadin±agarose beads. Puri®cation cysteine proteases involved in processes such as apoptosis and identi®cation of probe-bound proteins is the key step [12], cataract formation [37], and malarial infection [38]. for the application of chemical probes to proteomics (for a Other probes have been used to pro®le enzymes involved review of methods for protein identi®cation in proteomics in clinically relevant conditions such as cancer progres- see [32,33]). The biotin±streptavidin bond is one of the sion and cancer cell invasiveness [25,39]. In all of these strongest known non-covalent interactions with an asso- cases, complex disease states and proteomes were dis- ciation constant of 1015 MÀ1 [34] allowing for quantitative tilled into a few biochemically tractable enzymes that binding of low abundance biotinylated enzymes. How- could then be studied in more detail. ever, this tight binding interaction comes at price, as it can be dif®cult to elute biotinylated proteins from streptavi- Chemical probes have the potential to rapidly increase the din resins without using harsh conditions that result in the number of new drug targets. Recent studies have esti- elution of non-speci®cally bound and endogenously bio- mated that of the 483 known drug targets, only 122 have tinylated proteins. For this reason, a major focus in probe been targeted by orally available small-molecule inhibi- design in the future will be the inclusion of cleavable tors that are marketed to treat human disease [40,41]. linkers between the reactive group and the biotin tag that Almost half of these are enzymes that have the potential allow mild and selective elution of the probe-bound to be targeted with chemical probes. These studies also proteins from the solid support. A photocleavable linker estimate that there are ten times as many possible drug has been reported for use with ICAT probes [35] and it is targets remaining to be discovered. Chemical proteomics anticipated that other types of cleavable linkers will also has the potential to uncover these targets in a rapid, be developed. systematic and comprehensive manner through design, synthesis, and application of relevant probes. Applications of chemical proteomics Chemical probes can be used to study all aspects of Among published chemical probes, those designed to proteomics from protein expression and identi®cation target the cysteine protease family may have limited www.current-opinion.com Current Opinion in Biotechnology 2003, 14:87±95 92 Analytical biotechnology

potential for new target identi®cation given the small size probe can be useful to study protein phosphatases in a of this enzyme family in humans. By contrast, chemical complex setting. probes designed to target serine [42] hold a greater potential given that approximately 1% of the Applications to drug screening and drug ef®cacy human genome is predicted to encode members of this studies enzyme family [2,3]. Furthermore, a generally reactive The process of new target identi®cation often provides probe containing a sulfonate ester electrophile was found little information regarding the function of potential tar- to be useful for identifying abundant enzymes that are not gets in a given disease state. One strategy for target sequence related but might share common features in validation is to quickly identify lead compounds that substrate binding or catalytic mechanism [25]. It will be inhibit the target and use them to assess the functional interesting to see how pro®ling of these mechanistically role of the target in a disease model. However, identi®ca- related enzymes can be used to identify drug targets and tion of these lead compounds can be costly and time- increase our biological understanding of disease. consuming. Identifying small-molecule inhibitors of enzy- matic drug targets usually involves designing a high- In addition to the increasing number of reported chemical throughput assay that can be automated and used to screen probes, many mechanism-based inhibitors or af®nity large libraries of potential inhibitors. These standard enzy- labeling reagents have already been identi®ed that could matic screens often require large quantities of recombi- be used to develop new classes of chemical probes. Of the nantly expressed target proteins that can be dif®cult to major classes of enzymes that have been the focus of drug produce. In addition to the desired target, several closely discovery efforts, the two that are most amenable to related enzymes are often produced and screened to test chemical proteomics applications are protein kinases for drug selectivity. and protein phosphatases. These enzymes are critical regulators of cell signaling and metabolism and have Chemical probes have recently been shown to have great received much attention as possible drug targets [43,44]. potential to facilitate the process of lead drug identi®ca- tion (Figure 3a). DCG-04, a chemical probe directed The protein kinase family of enzymes is perhaps the most towards the papain family of proteases, was used to likely to bene®t from a chemical proteomic strategy. This identify preliminary selective inhibitors of individual family is made up of over 500 members with diverse cathepsin proteases [30]. A ¯uorescently tagged version substrate requirements and cellular functions [2,3]. For of DCG-04 was used in rat liver extracts in competition this reason, it is unlikely that any single chemical probe with a small library of inhibitors to identify compounds will be developed that will target the entire kinase family. that could selectively inhibit a desired target protease. Rather, many subclass-speci®c reagents will need to be This method does not require, a priori, knowledge of the identi®ed. Interestingly, no chemical probes for kinases, target enzyme, and screens performed in complex mix- other than 50-FSBA mentioned above, have appeared in tures provide information regarding the potency and the literature. However, covalent inhibitors of epidermal selectivity of each compound with respect to other closely growth factor receptor (EGFR) kinases and phosphatidy- related family members. However, initial screening linositol-3 (PI3) kinases have been identi®ed [45,46].Itis methods depend upon the use of gel-based separation not clear whether these reactive groups could in fact be techniques, allowing for only low-throughput applica- used to generate speci®c chemical probes for these tions. Further development will be required to make this families of kinases. approach amenable to high-throughput analysis. For example, the use of ¯uorescent probes with different Protein phosphatases are also emerging as desirable drug excitation and emission pro®les can be used in conjunc- targets as they are involved in many cell-signaling events tion with a high-throughput capillary electrophoresis sys- [47]. Phosphatases are particularly suitable for analysis tem, thereby allowing for unattended screening of with type I reactive groups on chemical probes because thousands to tens of thousands of compounds per day. the reaction mechanism involves a nucleophilic thio- late group on a cysteine residue in the active site [48]. Chemical probes could also be applied in drug ef®cacy In addition to the di-¯uoromethylphenols mentioned studies (Figure 3b). This process involves the identi®ca- above ([18,19]; Figure 2), haloacetophenones have been tion of the intended drug target in a complex proteome used as tyrosine phosphate mimetics to covalently modi- using a simple competition assay between the drug candi- fy phosphatases [49]. One problem with these probes is date and the chemical probe. The ef®cacy of the drug the low speci®city with which they bind phosphatases. candidate can be tested against the native enzyme in a With KI values in the range of 0.04 mM to >2 mM, they more physiologically relevant milieu than standard enzy- may not be speci®c enough to identify low abundance mological assays. Furthermore, the ability of a drug candi- phosphatases in a complex protein mixture. Speci®city date to inhibit a speci®c target and not other related targets elements in the linker region or perhaps even on the can be assessed within a relevant tissue sample. Such reactive group may be needed before these types of methods for evaluation of drug potency and selectivity

Current Opinion in Biotechnology 2003, 14:87±95 www.current-opinion.com Chemical proteomics in drug discovery Jeffery and Bogyo 93

Figure 3

(a) (b) Inhibitor: NO2 NO2 O NO2 S Inhibitor:

NH NH NH Off-targets Intended drug target Mouse model of disease Off-targets Off-targets NO2 Intended Dose mice with drug target inhibitor 1234 NH Fluorescence-scanned gel Off-targets Inject mice with chemical probe Remove tissue for gel analysis 1234 Fluorescence-scanned gel Current Opinion in Biotechnology

Applications of chemical probes to methods of drug discovery. (a) Use of a chemical probe in drug lead selection. Image of a fluorescence-scanned SDS±PAGE gel is shown depicting the reactivity of a fluorescently tagged chemical probe in a complex proteome (lane 1, red lines depict labeled enzymes). Hypothetical lead drug candidates are added in competition with the chemical probe (lanes 2±4). Enzyme family members whose labeling is inhibited by addition of a drug indicates active-site binding (dotted green circles). The drug lead in lane 2 shows the most specificity for the intended drug target. (b) Evaluation of drug efficacy in an animal model for disease using a chemical probe. In preclinical studies, an animal is treated with a lead drug candidate to assess efficacy. The animals can be injected with a fluorescently tagged chemical probe to evaluate the efficacy and selectivity of the drug towards the intended target in vivo. Lanes 2±4 of the fluorescence-scanned SDS±PAGE gel show that the drug candidate is indeed inhibiting the intended drug target (as measured by competition for labeling of the target by the fluorescent probe) in a dose-dependent manner and shows no activity towards related family members. using chemical probes could be carried out in a variety of speed with which quality drug candidates enter into the systems including puri®ed proteins, cells lysates, live cells, clinic. and even live animals. Information gained from such an analysis could help focus drug development efforts to The continued success of chemical proteomics depends increase potency and selectivity of a drug candidate within on the design of new probes and probe classes that can the context of a cellular or animal disease model. speci®cally target diverse sets of enzyme families. As chemical probes become more widespread in their use, Conclusions there will undoubtedly be other families of enzymes The sequencing of the human genome has had tremen- whose mechanism of action is particularly suited for dous impact on science and medicine. To maximize the investigation using chemical proteomic methods. bene®t from this rich resource of information, scientists must continue to advance the ®elds of genomics and Acknowledgements proteomics. The possibility of de®ning the function of We thank Amos Baruch for helpful discussions and critical reading of the every gene or protein in an entire genome now seems manuscript. possible, although it will require many years of work and will depend on continued technological innovation. References and recommended reading Papers of particular interest, published within the annual period of review, have been highlighted as: The ®eld of chemical proteomics represents the cross- roads of many disciplines that, when applied together, are  of special interest  of outstanding interest well suited to advance our understanding of biology and 1. Reiss T: Drug discovery of the future: the implications of the drug development. Without a doubt, the most important human genome project. Trends Biotechnol 2001, 19:496-499. tool of this ®eld is the chemical probe that must be 2. Lander ES, Linton LM, Birren B, Nusbaum C, Zody MC, Baldwin J, carefully designed to covalently attach to proteins of Devon K, Dewar K, Doyle M, FitzHugh W et al.: Initial sequencing interest and allow puri®cation and/or identi®cation. and analysis of the human genome. Nature 2001, 409:860-921. The chemical probes that have been developed to date 3. Venter JC, Adams MD, Myers EW, Li PW, Mural RJ, Sutton GG, have been shown to be useful for the identi®cation of new Smith HO, Yandell M, Evans CA, Holt RA et al.: The sequence of the human genome. Science 2001, 291:1304-1351. protease drug targets in a variety of different diseases. 4. Cockett M, Dracopoli N, Sigal E: Applied genomics: integration of They have also been used to rapidly identify drug leads. the technology within pharmaceutical research and These applications could have signi®cant impact on the development. Curr Opin Biotechnol 2000, 11:602-609. www.current-opinion.com Current Opinion in Biotechnology 2003, 14:87±95 94 Analytical biotechnology

5. Norin M, Sundstrom M: Structural proteomics: developments in Use of a sulfonate ester probe to identify enzymes involved in cancer structure-to-function predictions. Trends Biotechnol 2002, progression. This paper provides an example of using a general 20:79-84. electrophile to identify enzymes with a common mechanism of action rather than tailoring the reactive group to the intended enzyme 6. Pandey A, Mann M: Proteomics to study genes and genomes. family. Nature 2000, 405:837-846. 26. Borodovsky A, Kessler BM, Casagrande R, Overkleeft HS, 7. Cravatt BF, Sorensen EJ: Chemical strategies for the global Wilkinson KD, Ploegh HL: A novel active site-directed probe analysis of protein function. Curr Opin Chem Biol 2000, speci®c for deubiquitylating enzymes reveals proteasome 4:663-668. association of USP14. EMBO J 2001, 20:5187-5196. 8. Bogyo M, Verhelst S, Bellingard-Dubouchaud V, Toba S, 27. Borodovsky A, Ovaa H, Nagamalleswari K, Tudeviin G, Wilkinson Greenbaum D: Selective targeting of lysosomal cysteine  KD, Ploegh H, Kessler BM: Chemistry-based functional proteases with radiolabeled electrophilic substrate analogs. proteomics reveals novel members of the deubiquitinating Chem Biol 2000, 7:27-38. enzyme family. Chem Biol 2002, 9:1149-1159. This paper outlines a self-splicing protein (intein) approach to generate 9. Bogyo M, McMaster JS, Gaczynska M, Tortorella D, Goldberg AL, protein-based chemical probes containing a set of diverse electrophilic Ploegh H: Covalent modi®cation of the active site threonine of reactive groups. These probes were used to identify 13 known and 10 proteasomal b subunits and the Escherichia coli homolog HslV novel de-ubiquinating enzymes. by a new class of inhibitors. Proc Natl Acad Sci USA 1997, 94:6629-6634. 28. Amersham Bioscience: URL: http:// www5.amershambiosciences.com/aptrix/upp01077.nsf/Content/ 10. Liu Y, Patricelli MP, Cravatt BF: Activity-based protein pro®ling: gel_blot_typhoon_home_page the serine hydrolases. Proc Natl Acad Sci USA 1999, 96:14694-14699. 29. MolecularProbes: URL: http://www.probes.com/media/ publications/150.pdf 11. Thornberry NA, Peterson EP, Zhao JJ, Howard AD, Grif®n PR, Chapman KT: Inactivation of interleukin-1 b converting enzyme 30. Greenbaum D, Baruch A, Hayrapetian L, Darula Z, Burlingame A, by peptide (acyloxy)methyl ketones. Biochemistry 1994,  Medzihradszky KF, Bogyo M: Chemical approaches for 33:3934-3940. functionally probing the proteome. Mol Cell Proteomics 2002, 1:60-68. 12. Faleiro L, Kobayashi R, Fearnhead H, Lazebnik Y: Multiple species This paper outlines the potential utility of chemical probes in drug lead of CPP32 and Mch2 are the major active caspases present in identi®cation in a complex proteome. Screening of a small compound apoptotic cells. EMBO J 1997, 16:2271-2281. library against a set of mouse cathepsins present in a liver homogenate 13. Colman RF: Chemical arrows for enzymatic targets. FASEB J identi®ed speci®c inhibitors of cathepsin B. 1997, 11:217-226. 31. Patricelli MP, Giang DK, Stamp LM, Burbaum JJ: Direct  visualization of serine activities in complex 14. Walsh CT: Suicide substrates, mechanism-based enzyme proteomes using ¯uorescent active site-directed probes. inactivators: recent developments. Annu Rev Biochem 1984, Proteomics 2001, 1:1067-1071. 53:493-535. This paper highlights the advantages of using ¯uorescent tags on che- 15. Goshe MB, Smith RD: Stable isotope-coded proteomic mass mical probes. Advantages include sensitivity, throughput, and multiplex- spectrometry. Curr Opin Biotechnol 2002, 13:in press. ing of experiments through the use of ¯uorescent tags with non- overlapping excitation and emission spectra. 16. Greenbaum D, Medzihradszky KF, Burlingame A, Bogyo M: Epoxide electrophiles as activity-dependent cysteine protease 32. Rappsilber J, Mann M: What does it mean to identify a protein in pro®ling and discovery tools. Chem Biol 2000, 7:569-581.  proteomics? Trends Biochem Sci 2002, 27:74-78. A review focused on methods to identify a protein using current proteo- 17. Therrien C, Levesque RC: Molecular basis of antibiotic mics applications. resistance and beta-lactamase inhibition by mechanism-based inactivators: perspectives and future directions. FEMS 33. Gygi SP, Aebersold R: Mass spectrometry and proteomics. Microbiol Rev 2000, 24:251-262. Curr Opin Chem Biol 2000, 4:489-494. 18. Betley JR, Cesaro-Tadic S, Mekhal®a A, Rickard JH, Denham H, 34. Wilchek M, Bayer EA: Introduction to avidin-biotin technology. Partridge LJ, Pluckthun A, Blackburn GM: Direct screening for Methods Enzymol 1990, 184:5-13. activity by turnover-based capture of protein 35. Zhou H, Ranish JA, Watts JD, Aebersold R: Quantitative proteome catalysts. Angew Chem Int Ed Engl 2002, 41:775-777.  analysis by solid-phase isotope tagging and mass 19. Wang Q, Dechert U, Jirik F, Withers SG: Suicide inactivation of spectrometry. Nat Biotechnol 2002, 20:512-515. human prostatic and a phosphotyrosine An example of a cleavable linker applied to a chemical probe. Use of a phosphatase. Biochem Biophys Res Commun 1994, photocleavable linker and an isotope-coded af®nity tag (ICAT) probe 200:577-583. resulted in increased sensitivity over standard methods. 20. Janda KD, Lo LC, Lo CH, Sim MM, Wang R, Wong CH, Lerner RA: 36. Myers S, Baker A: Drug discovery ± an operating model for a new Chemical selection for catalysis in combinatorial antibody era. Nat Biotechnol 2001, 19:727-730. libraries. Science 1997, 275:945-948. 37. Baruch A, Greenbaum D, Levy ET, Nielsen PA, Gilula NB, Kumar 21. Colman RF: Af®nity labeling of purine nucleotide sites in NM, Bogyo M: De®ning a link between gap junction proteins. Annu Rev Biochem 1983, 52:67-91. communication, proteolysis, and cataract formation. J Biol Chem 2001, 276:28999-29006. 22. Zoller MJ, Nelson NC, Taylor SS: Af®nity labeling of cAMP-dependent protein kinase with p-¯uorosulfonylbenzoyl 38. Greenbaum DC, Baruch A, Grainger M, Bozdech Z, adenosine. Covalent modi®cation of lysine 71. J Biol Chem  Medzihradszky KF, Engel J, Holder AA, De Risi J, Bogyo M: 1981, 256:10837-10842. Identi®cation of a protease-dependent invasion pathway in the human malarial parasite, Plasmodium falciparum. Science 23. Hohenegger M, Freissmuth M, Nanoff C: Covalent modi®cation of 2002, in press. G proteins by af®nity labeling. Methods Mol Biol 1997, An example of the use of chemical probes to identify a new drug target in 83:179-190. a very important human pathogen, and to quickly screen for potent and speci®c inhibitors of the target. 24. Gygi SP, Rist B, Gerber SA, Turecek F, Gelb MH, Aebersold R: Quantitative analysis of complex protein mixtures using 39. Jessani N, Liu Y, Humphrey M, Cravatt BF: Enzyme activity isotope-coded af®nity tags. Nat Biotechnol 1999, 17:994-999.  pro®les of the secreted and membrane proteome that depict cancer cell invasiveness. Proc Natl Acad Sci USA 2002, 25. Adam GC, Sorensen EJ, Cravatt BF: Proteomic pro®ling of 99:10335-10340.  mechanistically distinct enzyme classes using a common An example of the use of a chemical probe to identify potential drug chemotype. Nat Biotechnol 2002, 20:805-809. targets and enzymatic diagnostic markers in cancer.

Current Opinion in Biotechnology 2003, 14:87±95 www.current-opinion.com Chemical proteomics in drug discovery Jeffery and Bogyo 95

40. Hopkins AL, Groom CR: The druggable genome. Nat Rev Drug 46. Wymann MP, Bulgarelli-Leva G, Zvelebil MJ, Pirola L, Discov 2002, 1:727-730. Vanhaesebroeck B, Water®eld MD, Panayotou G: Wortmannin inactivates phosphoinositide 3-kinase by covalent 41. Drews J: Drug discovery: a historical perspective. Science 2000, modi®cation of Lys-802, a residue involved in the phosphate 287:1960-1964. transfer reaction. Mol Cell Biol 1996, 16:1722-1733.

42. Kidd D, Liu Y, Cravatt BF: Pro®ling serine hydrolase activities in 47. Mustelin T, Feng GS, Bottini N, Alonso A, Kholod N, Birle D, Merlo J, complex proteomes. Biochemistry 2001, 40:4005-4015. Huynh H: Protein tyrosine phosphatases. Front Biosci 2002, 43. McCluskey A, Sim AT, Sakoff JA: Serine-threonine protein 7:85-142. phosphatase inhibitors: development of potential therapeutic 48. Denu JM, Dixon JE: Protein tyrosine phosphatases: strategies. J Med Chem 2002, 45:1151-1175. mechanisms of catalysis and regulation. Curr Opin Chem Biol 44. Cohen P: Protein kinases ± the major drug targets of the 1998, 2:633-641. twenty-®rst century? Nat Rev Drug Discov 2002, 1:309-315. 49. Arabcai G, Guo X, Beebe KD, Coggeshall KM, Pei D: 45. Garcia-Echeverria C, Traxler P, Evans DB: ATP site-directed a-Haloacetophenone derivatives as photoreversible covalent competitive and irreversible inhibitors of protein kinases. inhibitors of protein tyrosine phosphatases. J Am Chem Soc Med Res Rev 2000, 20:28-57. 1999, 121:5085-5086.

www.current-opinion.com Current Opinion in Biotechnology 2003, 14:87±95